RECOMMENDATION ITU-R BO.786 *

Transcription

1 Rec. ITU-R BO.786 RECOMMENDATION ITU-R BO.786 * MUSE ** system for HDTV broadcasting-satellite services (Question ITU-R /) (992) The ITU Radiocommunication Assembly, considering a) that the MUSE system has been developed for HDTV satellite broadcasting in the 2 GHz band; b) that the MUSE system has been tested for many years using a broadcasting satellite in operation; c) that there exists a studio standard based on 25 lines and 6 Hz field rate, which is very well coupled to the MUSE system; d) that a large number of programmes have been produced worldwide using the studio standard with 25 lines and 6 Hz field rate; e) that a down converter from the MUSE system to the M-NTSC system has been developed and is operating in conventional television sets; f) that the MUSE system can be used for terrestrial broadcasting, cable TV (CATV) distribution, and recording media, e.g. optical discs, VCRs, recommends that, for an administration or organization wishing to initiate a MUSE * 25 line, 6 Hz field rate based HDTV broadcasting-satellite service, the signals should conform to the specification contained in Annex. ANNEX Signal specification of MUSE system Introduction The MUSE system has been developed for HDTV satellite broadcasting in the planned 2 GHz band. Picture quality and system feasibility have been confirmed through various tests including daily operation using the BSS. The major signal specification of the MUSE system is described in this Annex. * Radiocommunication Study Group 6 made editorial amendments to this Recommendation in 2 in accordance with Resolution ITU-R 44. ** The MUSE (Multiple sub-nyquist Sampling Encoding) system is described in Annex.

2 2 Rec. ITU-R BO Technical bases for MUSE system The major technical bases employed for the MUSE satellite emission system are listed below: a bandwidth compression technique using multiple sub-sampling and motion compensation. This compresses the HDTV signal bandwidth of 3 MHz to 8. MHz suitable for HDTV satellite broadcasting in the 2 GHz band; time division multiplexing of luminance Y and chrominance C signals. This makes a complete Y/C separation possible in the decoder, resulting in the elimination of interference, such as cross colour and cross luminance which often appear in the existing TV systems; a technique for automatic waveform equalization required for analogue sampled-value transmission. This enables the system to be used not only in the case of satellite emission but also in other transmission media, such as CATV; a synchronization system maintaining accurate re-sampling phase in the decoder. The system employs a positive synchronizing signal, which gives the received picture a signalto-noise ratio (S/N) 3 db higher than that of the conventional synchronization system; an efficient non-linear emphasis system suitable for satellite transmission. This gives an emphasis gain or S/N improvement of 9.5 db; coding based on the principle of quasi-constant luminance. This not only reduces remarkably the cross-talk between the Y and C signals due to a narrower bandwidth limitation of the C signal, but also improves the S/N in highly saturated colour pictures. It leads to a lower carrier-to-noise ratio (C/N) required for BSS direct reception; a baseband multiplexing of digital sound and independent data. This allows a flexible usage of the system, as it is independent of the modulation system and transmission media. 3 MUSE signal specification The definitions given are applicable to all parameters and terms specified in this Recommendation: tolerance is zero for the parameter values with digital definition; the values that express the sampling frequency, bit rate, and transmission rate have the frequency relationship with the line frequency ( f L) of the input signal, as given in Table ; the video signal value is represented by an 8-bit digital expression, unless otherwise noted; the grey level of the video signal is defined by value 28 for the luminance signal; the levels for the luminance signal corresponding to black level, grey level and white clip level are defined by the signal values immediately after the inverse gamma correction and white clipping processes; the signal level of the guard area is equal to the grey level; undefined binary data in this specification will be filled in with either or.

3 Rec. ITU-R BO Transmission signal format Figure shows the transmission signal format. FIGURE Transmission signal format Clock No. (Clock: 6.2 MHz) Header Sound/data (4 + 4 lines) VITS No. Frame pulse No. VITS No. 2 Frame pulse No C signal (56 lines) Guard area YM signal (56 lines) ~ ~ ~ ~ ~ Line No Transmission control signal (5 lines) Clamping level Control signals of programme transmission Sound/data (4 + 4 lines) 69 C signal (56 lines) Guard area YM signal (56 lines) ~ ~ ~ ~ ~ 2 25 Transmission control signal (5 lines) VITS: vertical interval test signal Clamping level Note The transmission control signal is valid for the next field following the field that contains it. Note 2 Line No. 564 is assigned to uses such as control signals of programme transmission by broadcasters. Note 3 Timing relationship with the studio video signal is as follows: C signal in line No. 43 and YM signal in line No. 47 correspond to the studio signals in line No. 42. D

4 4 Rec. ITU-R BO Video transmission signal 3.2. Formulation of transmission signal A luminance signal, YM, as specified below is used so that the inverse matrix in a receiver can be made simple. YM =.588 G +.8 B R The matrix for the receiver is expressed by the following equation: G = B R /5 /2 5/4 B R YM YM YM = /4 5/4 /2 B R YM YM YM TABLE Relationship of frequencies used in the MUSE system with the line frequency Frequency 97.2 MHz = f L MHz = f L MHz = f L MHz = f L MHz = f L MHz = f L MHz = f L MHz = f L MHz = f L MHz = f L MHz = f L MHz = f L 2.35 MHz = f L 4 28 khz = (f L/ 25) 6 (8/3) 2 khz = (f L/ 25) 6 (7/3) 48 khz = (f L/ 25) 6 32 khz = (f L/ 25) 6 (2/3) khz = (f L/ 25) 6 (/48) where f L = Hz.

5 Rec. ITU-R BO Multiplexing of chrominance signal A line sequential time compressed integration (TCI) signal is employed for the multiplexing, where the chrominance signal is time-compressed to /4 and is time-division multiplexed with the luminance signal YM. The 48 samples per line are assigned to the luminance and chrominance signals as follows: Header (HD) C signal Guard area YM signal D The R-YM signal is multiplexed into odd lines and B-YM into even lines, respectively. (The C signal is advanced against the YM signal by 4 lines.) Relationship of active line periods between studio and transmission signals Figure 2 shows the relationship of active line periods between the studio and transmission signals. The active line period of the transmission signal corresponds to 22 sample duration at the sampling frequency of MHz. FIGURE 2 Relationship of active line periods between the studio and transmission signals Active transmission samples per line Active samples per line in studio standard Note The number of samples denoted here is counted by the clock of MHz used for the studio sampling frequency. D Phase relationship of YM and C signals relative to the original sampling lattice Figure 3 shows the sampling points of transmission signals, YM and C, in relation to the original sampling lattice.

6 6 Rec. ITU-R BO.786 D4

7 Rec. ITU-R BO Video encoding Multiple sub-nyquist sampling with inter-field and inter-frame offset is used for video encoding. Figure 4 shows the sub-sampling sequence in the encoder. Motion compensation is applied to the YM signal. In signal processing of the stationary picture area, from the output of field offset sub-sampling in the encoder to the input of inter-field interpolation in the decoder, the overall frequency characteristics should satisfy the distortion-free condition required for sampled-value transmission. The frequency characteristics after the field-offset sub-sampling process of the encoder should be such that they give the overall characteristics of a linear-phase roll-off with attenuation of 6 db at 2.5 MHz, when the reference filter having the impulse response defined in Table 2 is used. Note that the values in the table represent the impulse response at the sampling frequency of 97.2 MHz, and that the values are symmetric with respect to the centre tap (Tap No. ). The amplitude gain of the filter is set to 3 at d.c. FIGURE 4 Sub-sampling sequence in encoder Time compression Field-offset sub-sampling Sampling frequency conversion Frame-offset sub-sampling A B C D E Sampling frequencies (MHz): A: C: E: B: D: a) Sampling sequence for luminance signal Time compression Field-offset sub-sampling Frame-offset sub-sampling F G H J Sampling frequencies (MHz): F: H: G: J: b) Sampling sequence for chrominance signal D Transmission sampling frequency 6.2 MHz is used for the transmission sampling frequency.

8 8 Rec. ITU-R BO Video signal levels YM signal black level: 6 white clip level: 239 C signal achromatic level: 28 In the encoder, the amplitude of the C signal is increased by 3 db, followed by clipping at the level of either 6 or 239. TABLE 2 Filter impulse response of reference decoder Tap No. Response Quasi-constant luminance processing The quasi-constant luminance principle is applied to reduce the YM and C signal interference. The block diagram of signal processing and the characteristic curves for gamma correction are given in Figs. 5 and 6 to 8, respectively Inverse gamma curve for R, G and B signals The curve is expressed by the equation below, where the maximum value of quantization is normalized to unity and black level of the video signal is assumed to be zero. y = A (x B) g + C where g = 2.2 (both coordinates (,) and (,) belong to the curve and the ratio of gradient at (,) and (,) is :5).

9 Rec. ITU-R BO Transmission gamma curve for YM signal The curve resulting from the following expression is parabolic, where the maximum value of quantization is normalized to unity, and the black level of the video signal is assumed to be zero. Processing of the gamma signal should be applied only during the YM signal period. x = (3/5) y 2 + (2/5) y FIGURE 5 Block diagram of signal processing based on quasi-constant luminance principle R, G, B input Low pass filter A/D conversion Inverse gamma correction Transmission matrix Transmission gamma (for C signal) Sub-sampling, etc. (MUSE encoding) Transmission gamma (for YM signal) MUSE output D6

10 Rec. ITU-R BO.786 FIGURE 6 Inverse gamma curve for R, G and B signals Output level (normalized) y x Input level (normalized) D7 FIGURE 7 Transmission gamma curve for YM signal Output level (normalized) y x Input level (normalized) D8

11 Rec. ITU-R BO.786 FIGURE 8 Transmission gamma curve for C signal C Output level (normalized) y A B 24 8 x Input level (normalized) Transmission gamma curve for C signal The curve shown in Fig. 8 is for positive values of x, while it has origin symmetry for negative values. The curve is expressed by the following equations, where half of the full quantization range is normalized to unity. This transmission gamma process should be applied only during the C signal period. Between and A: y = (5/3) x Between A and B: y = (48/) x 2 + (67/33) x (/32) Between B and C: y = (3/33) x + (2/33) Synchronizing signal The positive sync with digital frame pulses is employed. The waveforms are shown in Figs. 9 and. The frame pulses are inserted in the latter half of lines No. and No. 2 respectively, as shown in Fig. 9.

12 2 Rec. ITU-R BO.786 D-sc 3.2. Others Clamping level signal The signal level of samples Nos. 7 to 48 in both lines No. 563 and No. 25 takes the value VIT signals The following signals are inserted at sample No. 264 in lines Nos. and 2: line No. : negative impulse with the duration of clock of 32.4 MHz; line No. 2: positive impulse with the duration of clock of 32.4 MHz.

13 Rec. ITU-R BO D-sc Transmission control signals The signals consist of 32 bit data, of which the information is listed in Table 3. The 32 bits are divided and arranged into eight groups of 4 bits, in incremental order. Each group consists of an 8-bit word with the (8,4) extended Hamming error protection code. The four information bits are allocated on the left half of the word, while four check bits are on the right half. The bits are transmitted from left to right. Generator matrix for the extended Hamming code is defined as follows: Each bit is transmitted in two transmission clock periods in NRZ form. The levels of the NRZ signal are as follows: logical : value 239 logical : value 6

14 4 Rec. ITU-R BO.786 TABLE 3 Transmission control signal Bit No. Parameter Description b b 2 b 3 b 4 b 5 b 6 b 7 b 8 b 9 b b b 2 Field offset subsampling phase for YM Horizontal motion vector Vertical motion vector Frame offset subsampling phase for YM Frame offset subsampling phase for C Control of noise reducer in the decoder Set to, when sampling point is in the right hand position The values are expressed in 2s complement, and set to positive when motion is to the right. The motion estimation accuracy corresponds to clock period of 32.4 MHz. Bit denoted by b 2 represents LSB The values are expressed in 2s complement, and set to positive when motion is downward. The motion estimation accuracy corresponds to one line spacing. Bit denoted by b 6 represents LSB Set to, when sampling points of odd lines are in the right hand position Set to, when sampling points are in the right hand position and when the quotient of line number divided by 2 is odd Four levels of noise reduction will be performed in the decoder according to the values expressed by the two bits (b is LSB). The reduction is larger with the higher value b 3 Equalization flag The bit is set to when equalization is in progress at the encoder b 4 Sensitivity control of motion detection Set to when low sensitivity b 5 b 6 b 7 b 8 Inhibit flag for two-frame motion detection Motion information b 9 Receiving mode control : standard mode : non-standard mode b 2 Modulation mode : FM : AM b 2 b 22 b 23 Extension bits When, two-frame motion detection is inhibited in the decoder The values expressed by the three bits (b 6 corresponding to LSB) show the following conditions: : normal : complete still picture 2: slightly in motion 3: scene change 4 to 7: represents the extent of motion Undefined b 24 Flag for still picture : normal mode : still picture mode b 25 to b 32 Extension bits Undefined, but are used for control of still picture application when b 24 is 3.3 Sound/data signal 3.3. Sound encoding Mode A and Mode B are provided for the transmission of the sound; a four-channel sound signal with 5 khz bandwidth for Mode A, and a 2-channel signal with 2 khz for Mode B.

15 Rec. ITU-R BO Mode A Mode B Signal bandwidth (khz) 5 2 Sampling frequency (khz) Quantization and companding for the sound signal Quantization and companding of the sound signal are described in Table 4. The differential value between two successive samples of the linearly quantized PCM signal is compressed nearly instantaneously. The sound signal transmission data are expressed in 2s complement and are sent from MSB to LSB. The bit length of each sample is 8 bits in the case of Mode A, and bits in the case of Mode B. TABLE 4 Quantization and companding method for sound signal Mode A Mode B Quantization 5 bit linear 6 bit linear Coding 2s complement Companding method Near instantaneous companding DPCM Companding law 5 to 8 bits with 8 ranges 6 to bits with 6 ranges Range bits The construction of the range bits for each single sound channel is shown below. The range data comprised of 3 bits is valid for the sound signal being transmitted in the same sound/data frame. Table 5 shows the relationship between compression ratios and the corresponding range data. Sending order Range data Check bits * LSB MSB * Control flag for scrambling identification of sound signal D2

16 6 Rec. ITU-R BO.786 TABLE 5 Relationship between compression ratios and range data Range data Compression ratio MSB LSB Mode A Mode B /28 /32 /64 /6 /32 /8 /6 /4 /8 /2 /4 / /2 / Multiplexing of sound/data signal Multiplexing Method Baseband time-division multiplexing Signal format Ternary, NRZ Instantaneous baud rate 2.5 MBd Binary/ternary conversion is specified as below. The ternary coded word is transmitted from left to right. Three binary symbol Two ternary symbol Sound/data signals Bit rate.35 Mbit/s Number of sound channels 4 (Mode A) or 2 (Mode B) Capacity of data channel 28 kbit/s (Mode A) or 2 kbit/s (Mode B) Framing code 6 bits/frame () Control code 22 bits/frame Word interleave 6 samples Bit interleave 6 bits Error correction code Sound/data BCH (82,74) SEC-DED Range data BCH (7,3) SEC-DED in addition to the above

17 Rec. ITU-R BO Generator polynomials for error correction code Sound/data (x 8 + x 7 + x 4 + x 3 + x + ) Range data (x 4 + x 3 + x 2 + ) Sound/data frame frequency khz is employed for the sound/data frame frequency Sound/data frame format Figure depicts the frame formats of Mode A and Mode B for the sound/data signals. FIGURE Sound/data frame format 35 bits Sound channel Sound channel 2 Sound channel 3 Sound channel 4 Check bits Range bits (for channels and 2) Control code Framing code Range bits (for channels 3 and 4) a) Mode A Independent data 35 bits Sound channel Sound channel 2 Check bits Range bits (for channels and 2) Control code Framing code b) Mode B Independent data D Data packet A packet multiplexing system is used for data transmission. The packet format is shown in Fig. 2.

18 8 Rec. ITU-R BO.786 FIGURE 2 Data packet format 288 bits Header Data a) Packet format 6 bits 5 SI Check bits SI: service identification code b) Header format D4 4 Conditional access Line rotation, line permutation and a combination of these are applicable as scrambling methods for the MUSE video signal. The access control methods of the MUSE system are quite similar to those of the digital subcarrier/ntsc system, which is described in Chapter of the ITU-R special publication Specifications of Transmission Systems for the Broadcasting-Satellite Service. 5 Modulation Frequency modulation is used for the BSS. The major parameter values are listed below. 5. Type of modulation Frequency modulation with keyed AFC (Level 28 corresponds to the centre frequency of the channel). 5.2 Modulation polarity Positive polarity. 5.3 Channel bandwidth 27 MHz or 24 MHz.

19 Rec. ITU-R BO Baseband signal bandwidth 8. MHz (with % root-cosine roll-off characteristics). 5.5 Energy dispersal signal Symmetric triangle waveform of 3 Hz (see Fig. 3). The frequency deviation for energy dispersal is set to 6 khz peak-to-peak. FIGURE 3 Energy dispersal signal Start point of line No. Middle point of line No. 563 Frequency deviation 6 khz 2nd field st field D5 5.6 Frequency deviation Video signal:.2 ±.5 MHz peak-to-peak (for 27 MHz channel bandwidth) 9. ±.5 MHz peak-to-peak (for 24 MHz channel bandwidth), where the corresponding signal levels are values 6 (black level) and 239 (white clip level). Ternary signal 8.6 ±.5 MHz peak-to-peak (for 24 MHz channel bandwidth) for sound/data: 9.8 ±.5 MHz peak-to-peak (for 27 MHz channel bandwidth). 5.7 Emphasis The emphasis characteristics E( f ) are derived from those of the de-emphasis D( f ) below, as E( f ) = /D( f ). D( f ) = (/2) + (5/6) cos (2π f / f s ) + (/8) cos (4π f / f s ) + (/6) cos (6π f / f s ) where: f s = 6.2 MHz. Emphasis gain = 9.5 db.

20 2 Rec. ITU-R BO.786 Non-linear characteristics: see Fig. 4. The emphasis is applied during the following periods: video signal periods (YM and C); those samples, among the samples No. and No. in the header (HD) period, that are adjacent to video signals; guard area (sample No. 6). FIGURE 4 Non-linear characteristics for emphasis Output signal level A Ellipse curve B Note The curve in the figure shows only positive values. It has origin symmetry for negative values. Note 2 The curve is specified as follows: between and A: straight line with gradient unity, between A and B: ellipse curve with gradient of /5 at point B. Note 3 In the figure, the level expression is different from that defined previously. The level in the figure corresponds to the grey level of the signal, while the level 224 in the figure corresponds to the white clip level. D6